What this is
Voices of Bumpy is a project that generates full songs across every conceivable genre of music, using cat noises as the vocals. The cat noises are written as lyrics, fed to Suno, and the model generates the full track.
That part is the setup. The joke is that Suno takes the material completely seriously. It delivers prog rock with full conviction. It does ambient with genuine atmosphere. It commits to the bit with an earnestness that makes it funny.
But the real art is in what Minimax M2.7 generates alongside the cat-noise lyrics. When given the premise with a genre prompt, it spontaneously produces commentary — explaining the history of the subgenre, why the genre is structured the way it is, what makes it gimmicky, or what to expect when pushing into weirder styles. It picks up on the absurdity and runs with it, generating ironic analysis that makes you understand both the genre and why this project works.
Each piece taken alone is mediocre. The full cycle — cat noises + earnest genre delivery + spontaneous AI commentary — is the art.
The name
“Bumpy” is our cat. He provides the majority of the source material in spirit. I’m still working on recording his delightfully melodic meows. Soon.
On the counterargument
Yes, this is a gimmick. That’s the point. Musicians already do different genres with unconventional vocals. People already discuss microgenres. People already generate AI music. Voices of Bumpy is the entire cycle, each step taken at face value and played completely straight. The comedy and the commentary only emerge from the whole. What I actually find hilarious about this is that I just had to tell it “Come up with more genres” and it made these. These were generated through Minimax M2.7, and it latched onto the bit and kept on trucking. The tips are things it took into account when creating the style prompts and lyrics, and some of them are genuinely inspired. All of the bits taken as a whole represent what one can argue is art, while none of the individual components could be, part of what makes it interesting is honestly the fact that it’s all the AI’s interpretation, and in some cases interpretation of interpretation. Alone each part of the process is just AI stuff. Then there’s also the arguement of whether this constitutes genuine creativity, or does the credit (or blame) go to me? I honestly don’t care about the answer personally. It’s meta as hell, and that’s the point.
All tracks are works in progress, some of them I might go back and revise so they’re a better example of the genre. The specific track isn’t the entire point.